53 lines
2.0 KiB
Markdown
53 lines
2.0 KiB
Markdown
# Assistant
|
|
|
|
Long-lived, evolvable personal/work assistant. This repository is intentionally designed for implementation primarily through Codex while preserving strict semantic, authority, storage and deployment boundaries.
|
|
|
|
## Fixed project coordinates
|
|
|
|
- **Expected local project folder:** `Z:\mst\codex\assistent`
|
|
- **Git remote:** `https://git.cin.su/m/Assistant.git`
|
|
- **Deployment control plane / Portainer:** `https://docker.cin.su`
|
|
- **Go baseline:** 1.26.x
|
|
- **PostgreSQL baseline:** 18.6
|
|
- **Semantic generation:** 1
|
|
- **Architecture generation:** 1
|
|
|
|
The repository at `git.cin.su/m/Assistant` was externally reachable and empty when this bootstrap pack was created. `docker.cin.su` was externally reachable as Portainer.
|
|
|
|
## First rule
|
|
|
|
Read `AGENTS.md`, then `architecture/ARCHITECTURE_CONSTITUTION.md`, then the current task packet under `dev/tasks/` **before editing code**.
|
|
|
|
## Bootstrap on Windows
|
|
|
|
```powershell
|
|
Set-Location Z:\mst\codex\assistent
|
|
powershell -ExecutionPolicy Bypass -File .\scripts\bootstrap.ps1
|
|
```
|
|
|
|
Then:
|
|
|
|
```powershell
|
|
go run .\cmd\devctl doctor
|
|
go run .\cmd\devctl check-architecture
|
|
go test ./...
|
|
docker compose -f deploy/docker-compose.dev.yml up -d --build
|
|
```
|
|
|
|
## Credentials
|
|
|
|
Do **not** commit passwords, API keys, tokens or copied credentials from sibling projects.
|
|
|
|
Codex should try, in this order:
|
|
|
|
1. existing OS/Git/Docker credential helpers without printing the secret;
|
|
2. existing environment variables / secret stores already configured for this machine;
|
|
3. inspect sibling projects only for non-secret deployment conventions, endpoint names and environment-variable names;
|
|
4. if actual credentials are still unavailable, ask the user.
|
|
|
|
See `docs/CREDENTIALS_AND_SECRETS.md`.
|
|
|
|
## V1 non-goals
|
|
|
|
V1 must not implement autonomous self-modification, global collective learning, a universal ontology engine, Kubernetes as a requirement, custom consensus, autonomous plugin installation, full robotics/control-loop infrastructure, or a multi-user federation platform. Preserve extension paths; do not materialize speculative systems.
|